Prakash Kishore Hazam1 · Gaurav Jerath1 · Nitin Chaudhary1 · Vibin Ramakrishnan1
Accepted: 10 July 2017
© Springer Science+Business Media, LLC 2017
utility of de novo designed peptides with diversified stereo- chemistry as a promising new approach in the generation of novel antibiotic peptides.
Keywords Peptidomimetics · Polymer tacticity · Antimicrobial peptides · Peptide design · Gramicidin · Stereochemistry
Abbreviations
MD Molecular dynamics RMSD Root mean square deviation Rg Radius of gyration
FE-SEM Field emission scanning electron microscopy MS Mass spectrometry
CFU Colony forming unit
Introduction
Higher organisms use peptides as a defense mechanism against invading pathogens (Dosler and Karaaslan 2014;
Ravichandran et al. 2017; Czihal and Hoffmann 2009;
Saikia et al. 2017), Defensins (Lehrer 2004), cecropins (Moore et al. 1996), magainins (Berkowitz et al. 1990) and cathelicidins (Zanetti 2005) constitute major classes of
“host defense peptide”. There is a remarkable increase in peptide based antibiotics, with nearly two dozen molecules are either completed or under clinical trials as antimicro- bial or immunomodulatory agents (Felicio et al. 2017).
However, cost, uncertain toxicology profiles and lability to proteolytic susceptibility hampers the otherwise prom- ising approach, especially in the light of the phenomenal increase in antibiotic resistance against conventional ther- apeutics (Li et al. 2015; Koh et al. 2015). Decreasing the size of peptides and machine learning approach have more Abstract Biocompatibility, low toxicity and high selec-
tivity towards bacterial cells has been the hallmark of peptide-based antibiotics. The innate immune system has been employing such molecular systems against invading pathogens as a successful defense strategy. In this study, we attempt to develop topologically constrained antimicrobial peptides with syndiotactic stereochemical arrangement, by incorporating L and D amino acids successively in its amino acid sequence. Acetylated versions of the designed peptides were also examined for its influence on bacteri- cidal potency, against Gram-positive and Gram-negative bacteria. Syndiotactic stereochemical arrangement of the polypeptide main chain mimics stereochemistry of Grami- cidin, a naturally occurring antimicrobial peptides. Gram- icidin is a class of penta-deca-peptides isolated from soil bacteria Bacillus brevis, but their utility as antibiotic was limited to topical use due to high levels of hemotoxicity.
Activity profiles of the four de novo designed peptide vari- ants show higher specificity towards Gram-positive bacte- ria than Gram-negative variants, matching earlier reports on the therapeutic potential of gramicidin as a broad spec- trum antibiotic. Significantly, our hemolytic assay confirms very low (<1%) levels of toxicity for the designed peptides unlike gramicidin. Earlier reports confirm that incorpora- tion of D amino acids effectively negates the possibility of proteolytic degradation, thus pointing to the potential Electronic supplementary material The online version of this article (doi:10.1007/s10989-017-9615-3) contains supplementary material, which is available to authorized users.
* Vibin Ramakrishnan [email protected]
1 Department of Biosciences and Bioengineering, Indian Institute of Technology Guwahati, Assam 781039, India
Author's personal copy
TH-2206_11610627
Int J Pept Res Ther or less addressed the first two issues, while the problem
of proteolytic degradation may be sorted out by the use of D-stereoisomers and unnatural amino acids in the sequence (Hamamoto et al. 2002). Designed antimicrobial peptides, in general, are cationic amphipathic peptide sequences with distinct membrane permeabilizing ability (Hazam et al.
2017; Wimley 2010). Though various mechanisms have been proposed like carpet (Tene et al. 2016), detergent (Bechinger and Lohner 2006), barrel stave (Chang et al.
2008), toroidal pore (Brogden 2005) etc., a consensus pic- ture is yet to emerge.
We designed two amphipathic peptide sequences and their N-terminal acetylated variants mimicking the ste- reochemical sequence of Gramicidin, a penta-deca-peptide isolated from soil bacterial species Bacillus brevis (Urry 1971). Antibiotic profile of Gramicidin reports that they are particularly effective against Gram-positive bacteria. How- ever, their utility is rather limited to topical use as a lotion or ointment in the treatment of infected surface wounds and throat infections (Wang and Ishii 2012; Wishart et al.
2008). Reported mechanism of action for Gramicidin sug- gests that it inserts itself into bacterial membranes resulting in membrane disruption and permeabilization, leading to a sequence of events such as the loss of intracellular sol- utes, dissipation of trans-membrane potential, inhibition of respiration and reduction in ATP pools, culminating in cell death (Burkhart et al. 1998). Naturally-occurring Gramici- din or Gramicidin D exists in isoforms with a variation of valine or isoleucine at first position. Gramicidin molecule is a mixture of Gramicidin A, B and C with a variation at the 11th position containing tryptophan, phenylalanine and tyrosine respectively (Burkhart et al. 1999). In this paper we present a design philosophy to synthesize cationic, amphip- athic peptide molecules incorporating D-amino acids in the design alphabet. To promote the membrane permeabilizing ability, we design our sequence around Gramicidin with gramicidin helix (6.3 helix) as the target structure. Grami- cidins form cation selective trans-membrane ion channel in typical model membranes (Hladky and Haydon 1972).
Even though, Gramicidin acquires various folding states, two of the major conformations have been predominant in different environments; first is the channel forming single stranded helical dimer and the second type is a non-chan- nel forming intertwined helix (Urry 1971; LoGrasso et al.
1988). Even though conformations of gramicidin are sol- vent dependent, single stranded β (6.3) dimer is the thermo- dynamically most stable structure (LoGrasso et al. 1988;
Killian et al. 1988; Kelkar and Chattopadhyay 2007).
Like all natural proteins, most of the therapeutic pep- tides are sequences of asymmetric amino acids having L-chiral stereo-chemical structure. In other words, they are stereo-regular “isotactic” hetero-polymers. The isotactic stereochemistry of the backbone, coupled with planarity
of peptide bond and steric effects limits the conformational space of a polypeptide. Even then, the predictive ability of a given amino acid sequence to fix the structure, folding intermediates and pathways are still weak, despite making enormous advancement in the basic understanding of this unsolved problem (Ramakrishnan et al. 2012; Khoury et al.
2014). This limits the success of a protein de novo design experiment in realizing its translational objectives.
Polymer sequences with alternating l and d-chiral monomers (amino acids) are referred to as syndiotactic polymers and with random distribution of l and d are het- erotactic or atactic polymer (Billmeyer 1957; Durani 2008).
Gramicidin is a 15 amino acid long trans membrane helical molecule with alternating l and d-chiral amino acids in its sequence and can hence be termed as a syndiotactic peptide (Nanda et al. 2007). In this study we tried to develop topo- logically constrained antimicrobial peptide molecules with syndiotactic stereo-chemical arrangements after making an objective assessment about their relative stability by molec- ular dynamics simulations. Acetylated variants of both the designed peptides, were also examined to test their possible differential efficacy, against Gram-positive and Gram-neg- ative bacterial species. The activity profiles were consider- ably similar, indicating limited significance of acetylation in syndiotactic sequences. The bactericidal potency of the designed peptides was higher against Gram-positive than in Gram-negative bacteria, though the design approach may be further modified in future designs in the generation of broad spectrum antibiotics.
Materials and Methods Design and Simulation
The peptides molecules used in this study were designed by using an in-house software PDB make and a modified version of Ribosome Software from G.D. Rose laboratory (Srinivasan 2013). Molecular dynamics (MD) simulations were performed using GROMACS (version: 4.6.5) (Pronk et al. 2013) with GROMOS 96 43a1 force field (Scott et al.
1999). The peptide structures were energy minimized in vacuum over 2000 steps using the steepest descent algo- rithm in a cubic box, with each face separated at 1.5 nm from the protein molecule. Subsequently, water molecules (SPC) were added to the simulation box and the system was further energy minimized. A 10 ns production run under NVT conditions was performed with an integration step of 2 fs. Initial velocities were taken from Maxwell distribu- tion at 300 K, using Berendsen thermostat for temperature coupling with coupling relaxation step setting at 0.1 ps.
The cut-off limit for non-bonded interactions was set at
Author's personal copy
Int J Pept Res Ther
1 3
0.8 nm. All through the simulation run, bond length was constrained with geometric accuracy of 10−4.
Electrostatic Profiling
Electrostatic potential for the designed peptides was cal- culated by solving Finite Difference Poisson–Boltzmann equation using Delphi software (Li et al. 2012). These elec- trostatic potential maps of candidate structures were further compared to identify distinct electrostatic potential signa- tures for each structure. The electrostatic potentials were represented in a 3D graph using MATLAB.
Reagents
Rink Amide resin, Fmoc amino acids, N, N, N, N-tetra- methyl-O-(1H-benzotriazol-1-yl) uronium hexafluo- rophosphate (HBTU), N, N-dimethylformamide (DMF), m-Cresol, Glutaraldehyde, Dimethylsulphoxide (DMSO), and Ethanol were purchased from Merck. 1-Hydroxy- benzotriazole (HOBt), diethyl ether, 4-(2-hydroxyethyl)- 1-piperazineethanesulfonic acid (HEPES), sodium phosphate monobasic anhydrous, acetonitrile (ACN), eth- ylene diamine tetra acetic acid (EDTA) and Piperidine were obtained from SRL laboratories. N, N-diisopropylethyl- amine (DIPEA), thioanisole, 1, 2 ethane dithiol (EDT), Tri- fluoroactic acid (TFA), 3-(4,5-dimethylthiazol-2-yl)-2,5-di- phenyltetrazolium bromide (MTT), was purchased from Sigma-Aldrich. Nutrient Broth and Agar was purchased from Himedia laboratories. All other reagents used were of the highest purity (≥99%).
Peptide Synthesis and Characterization
The peptides were synthesized employing standard Solid phase peptide synthesis protocol (Fmoc chemistry). Post synthesis, the peptides were cleaved from the resin using a cleavage cocktail (m-Cresol: Thioanisole: EDT: TFA ::
2:2:1:20) for 12 h in dark. The peptides were purified by means of semi-preparative reversed-phase liquid chroma- tography. A chromatographic run of 10% ACN to 100%
ACN with 0.1% TFA, was used for the gradient elution of peptides at a flow rate of 1 mL/min. The eluents were mon- itored at 210 nm and verified by electrospray ionization mass spectrometry (Wang et al. 2012; Falcao et al. 2016;
Chaudhary and Nagaraj 2011).
Antibacterial Assay
Staphylococcus aureus (NCTC 8530) and Pseudomonas aeruginosa (NCTC6750) bacterial strains were used for the assay. Mid logarithmic phase bacterial cells were washed and resuspended in sodium phosphate buffer (10 mM, pH
7.4). The resulting suspension was diluted to a net absorb- ance value of 0.2 at 600 nm and 50 μL of the inoculum was treated with required concentrations of peptide (Net incubation broth is 100 µL), incubated for 2 h. Post incu- bation, 100 µL of MTT solution (0.5 mg/mL) was added to the broth and incubated for 4 h at 37 °C. Subsequently, the reaction broth was centrifuged and the bacterial pellets were mixed with DMSO. The difference in the absorbance at 570 and 660 nm was calculated and percent bacterial cell lysis was reported, relative to untreated bacterial cells (Rufian-Henares and Morales 2011; Piaru et al. 2012; Eloff 1998). A relative inhibition of 80% growth with reference to growth control was considered to be as Minimal Inhibi- tory Concentration, as per relative reported standard proce- dure (Wu et al. 2014).
Field Emission-Scanning Electron Microscopy (FE-SEM)
Post incubation, glutaraldehyde (4%) was added to peptide- bacteria mixture, followed by an incubation of 30 min. The entire broth was centrifuged (1500×g, 5 min) and the bac- terial pellet was placed over a glass slide. The dried sam- ples were washed (gradient, 30–100% ethanol), coated with gold and images were recorded using scanning electron microscopy (Abd-El-Aziz et al. 2015).
Hemolytic Assay
Fresh human blood was collected and mixed with 1.5 mg EDTA per milliliter and mixed thoroughly. These red blood cells were re-suspended and repeatedly washed with buffer (5 mM HEPES buffer saline, pH 7.4) by centrifugation at 800×g for 5 min, until the supernatant was colorless.
A 10% cell suspension in buffer was prepared and 50 µL of the suspension was mixed with peptide solution, incu- bated for 2 h at 37 ± 2 °C. After incubation, the suspension was centrifuged at 800×g for 5 min and the absorbance of supernatant was recorded at 540 nm (Balaji and Trivedi 2012; Lee et al. 2014). Percent cell lysis was calculated in comparison to the extent of hemolysis recorded with 100%
lysis of blood cells in water.
Results and Discussion Design Philosophy
We designed two pairs of syndiotactic sequences mimick- ing Gramicidin 6.3 helix as model systems, with a starting (φ, ψ) dihedral angle pair shown in Table 1. Gramicidin has a right handed helical structure with (i to i + 7) and (i to i − 5) (Ramakrishnan et al. 2005) hydrogen bonding
Author's personal copy
TH-2206_11610627
Int J Pept Res Ther pattern, and a pitch 4.7 Å (Katsaras et al. 1992), with 6.3
residues per turn (Urry 1971). Theoretically such helical structures can assume two structural variants depending
on their ‘handedness’; either right or left handed (Fig. 1).
Therefore, we have designed two variants right handed (RH) and left handed (LH) for each sequence resulting in eight possible variants, though only four variants are syn- thetically possible.
The designed molecules are both cationic and amphi- pathic, which may play an important role in membrane lysis mediated by electrostatic interactions. Electrostatic potential maps of the designed peptides were generated by solving Finite Difference Poisson Boltzmann equation using Delphi software. The topology resulting from the side-chain orientation and the electrostatic potential envi- ronment it creates, were visibly distinct in all four models (Fig. 2). Precisely, the designed molecules display distinct topological and electrostatic fingerprint, which may get translated to their antibacterial potency.
The stereochemistry and sequences of 01 and 01-Ac pair and 02 and 02-Ac pair are the same, except that 01-Ac and 02-Ac being N-terminal acetylated (Table 1). To verify the stability of the designed peptides in a 6.3 helical conforma- tion, similar to gramicidin helix, a 10 ns MD run at 298 K under NVT condition using GROMOS 96 force field, was performed for each structural variant. The cluster analysis of the theoretical models suggest that the structures are quite stable with the percent sampling of most populous structure being more than 95% in all eight models (Supple- mentary Fig. 1; Supplementary Table 1).
Further, the structural integrity of the molecules was verified by measuring Root Mean Square Deviation (RMSD) from the starting structure (Fig. 3). Radius of Gyration has been measured to verify whether the designed Fig. 1 Backbone structures of the designed hetero-chiral peptides
having syndiotactic stereochemistry. All the peptides can either be in left or right handed orientation, resulting in a maximum of eight theo- retical backbone combinations possible for four peptides. a (01LH) and b (01RH) are the left handed and right handed back bone of sequence 01, whereas c (01Ac-LH) and d (01Ac-RH) are the left and right handed backbone of 01Ac. Likewise, e (02LH) and f (02RH) are left handed and right handed back bone of sequence 02 along with g (02Ac-LH) and h (02Ac-RH) are left and right handed version of 02Ac
Fig. 2 Electrostatic potential map of the designed peptides, using Delphi Software solv- ing Finite Difference Poisson Boltzmann equation. a, b are the potential maps of left and right handed conformations of the peptide 01. Similarly, c, d are the potential maps of the left handed and right handed peptide 02 respectively.
Though two conformations are theoretically possible, only one sequence can be experimentally verified. Differential color coding of amino acid residues, points to their electrostatic potential with reference to the color bar. The color map shows the potential distribution in KT/e units. (Color figure online)
Author's personal copy
Int J Pept Res Ther
1 3
structures remain folded throughout the simulation (Fig. 3).
Detailed examination of average structure forming largest cluster, however shows that hydrogen bonding pattern has been changed during the course of MD simulations (Sup- plementary Figs. 2 and 3; Supplementary Table 2). In our earlier studies, we have shown that gramicidin 6.3 helical structures with alternate l and d stereo-chemical sequence have their short-range and long range backbone electro- static interactions complementing each other, while in poly L sequences they are in conflict (Ramakrishnan et al. 2006;
Kumar et al. 2009; Ranbhor et al. 2006).
The resultant modification of energy landscape leading to extra stability due to complimentary local electrostat- ics is the key for stereo-chemical engineering of peptide sequences. We have further shown that countervailing short range–long range interaction holds the key for side-chain sequences dictating the overall conformation of a given polypeptide sequence. Since local electrostatics of gramici- din helical backbone is complimenting and stable, the fold is much more adaptable to varied sequence solutions (Ram- akrishnan et al. 2005). Our MD results on all eight vari- ants support earlier observations and this design strategy.
Since amphipathicity and cationicity has been the hallmark of antibacterial activity of natural and designed peptides, we have designed all peptides retaining amphipathicity, at the same time presenting differential electrostatic potential arising from cationic side-chains.
Antibacterial Assay and Toxicity Studies
The in vitro antimicrobial activity of the synthesised pep- tides was tested against Gram-positive (S. aureus) and Gram-negative (P. aeruginosa) bacteria. All the peptides were showing better activity against Gram-positive bac- teria, in comparison with Gram-negative bacteria (Fig. 4;
Table 1).
This observation is consistent with the earlier reports on gramicidin. Gramicidin has a reported MIC value of 2.2 µM (Xiong et al. 2005) against S. aureus and is inactive against Gram-negative bacteria. Our peptides show similar trend, though the sequence similarity between gramicidin and the designed peptides are minimal. Our designed pep- tides primarily rely on cationicity to induce membranolytic activity. Acetylation of synthesized peptides is aimed at lowering the net positive charge by blocking the free N-ter- minus. We, however also could not find any notable differ- ence in antimicrobial potency between acetylated and non- acetylated peptide variants.
Hemolytic Activity
Hemolytic assay against mammalian RBC was performed to estimate the extent of toxicity. All four peptides exhib- ited negligible cytotoxicity against mammalian cells at 100 µM concentration following standard protocols. Gram- icidin has a reported toxicity estimate of 50% hemolysis at 5 µM concentration (Wang et al. 2012). Unlike gramicidin, toxicity levels of all four reported peptides are well within the accepted safe threshold of 2% lysis at a very high con- centration of 100 µM; and three out of the four peptide var- iants are showing a value less than 1% (Table 2).
Microscopic Analysis
A qualitative evaluation of membrane rupturing poten- tial of designed antimicrobial peptide against bacteria was performed using FE-SEM analysis. The cationic peptides caused significant membrane damage which was inferred after a comparative analysis between peptide treated (rup- tured) and untreated (intact) bacterial cells. The bacte- rial cells were treated with 100 µM peptide resulting in deformed bacterial membrane architecture. (Fig. 5).
Table 1 Designed peptides and their sequences. The letter in the upper case denotes l-chiral amino acid whereas the letter in the lower case represents D. The sequences with N-terminal acetylation have been denoted with a prefix Ac. The dihedral angles used for the
peptide sequences has been shown in details. Minimum Inhibitory Concentration of the designed peptides against S. aureus and P. aer- uginosa shows preferential activity of syndiotactic sequences against Gram positive bacteria
Code Sequence Handedness Dihedral angles
(l-amino acid) Dihedral angles
(d-amino acid) Molecular mass
(Dalton) MIC (µM)
S. aureus P. aeruginosa
01 RkRaIvKrKvAa Right −120, 140 (φ, ψ) 120, −110 (φ, ψ) 1394.76 25 >100
Left −110, 120 (φ, ψ) 110, −150 (φ, ψ)
01 Ac Ac-RkRaIvKrKvAa Right −120, 140 (φ, ψ) 120, −110 (φ, ψ) 1436.8 25 >100
Left −110, 120 (φ, ψ) 110, −150 (φ, ψ)
02 kRkIaArArLvA Right −120, 140 (φ, ψ) 120, −110 (φ, ψ) 1351.69 12.5 >100
Left −110, 120 (φ, ψ) 110, −150 (φ, ψ)
02 Ac Ac-kRkIaArArLvA Right −120, 140 (φ, ψ) 120, −110 (φ, ψ) 1393.73 12.5 >100 Left −110, 120 (φ, ψ) 110, −150 (φ, ψ)
Author's personal copy
TH-2206_11610627
Int J Pept Res Ther
Qualitative pictures indicate that the membrane rup- turing ability of the amphipathic cationic peptides are responsible for the desired activity profile of the designed peptides, though detailed mechanism investiga- tion is beyond the scope of this paper.
Conclusions
Almost all natural proteins are isotactic with chain ste- reochemistry (poly L). An isotactic peptide molecule roughly acquires 21% allowed conformational space in Ramachandran plot. If the stereochemical sequence is a variable, it increases the designable space manifold, Fig. 3 a Root Mean Square Deviation (RMSD) as a function of time
of all the designed peptides from MD simulation in a 10 nanosecond (ns) production run, b Radius of Gyration (Rg) as a function of time indicating the overall folded nature of the designed peptides
Fig. 4 Bactericidal potency of the peptides 01 and 02 along with their acetylated variants against S. aureus (Gram-positive) and P. aer- uginosa (Gram-negative) bacteria. a and b show bactericidal activity against Gram-positive bacteria, whereas c and d show activity against Gram-negative bacterial species
Table 2 Hemolytic assay of designed peptides against mammalian Red Blood Cells (RBC) at 100 µM peptide concentration
S. no Peptide code % RBC lysis Standard
deviation
1 01 0.5 0.02
2 01 Ac 2 0.05
3 02 0.7 0.03
4 02 Ac 0.09 0.02